Goals: The goal is to expand the known pharmacological relationships among arrestin and G protein biased ligands. Starting with the Gi-coupled dopamine D2 receptor (D2R), we will predict biased ligands, evaluating our methods both retrospectively and via prospective experimental assay. We then extend this work to additional receptors, and finally organize the results into a map relating receptor polypharmacology by the functional selectivity of their ligands. Significance. Whereas biased ligands that preferentially activate arrestin or canonical signaling may comprise novel therapeutic opportunities unexploited among the well trammeled G protein coupled receptors, their discovery is more often serendipitous and few if any predictive computational models exist for it. The field suffers from a paucity of actual biased ligands, whose incompletely explored ranks include even marketed drugs such as aripiprazole, which may owe its efficacy to this mechanism. New screening capabilities to measure ?-arrestin1/2 recruitment at large scale provide unique and timely support for the computational methods, experimentally confirmed biased ligands, and biased polypharmacology maps proposed here. Theory/Background: Not all ligands are created equal. For GPCRs, ligands similar in selectivity, polypharmacology, and pharmacokinetics nonetheless exhibit differing patterns of second messenger signaling and therapeutic efficacy. This functional selectivity differentiates ligands that activate signal transduction via the canonical (G coupled) and the arrestin1/2 (aka arrestin3/4, hereafter arrestin) pathways. Pathway engagement is ligand specific and contributes to drug outcomes. As one example, effective third generation antipsychotics that induce fewer side effects were originally thought to be partial agonists of D2R, but may instead signal exclusively via arrestin2. Aim 1: To predict and test new arrestin-biased ligands. Biased ligands are thought to stabilize transmembrane receptor conformational ensembles to preferentially activate a signaling pathway; in practice however we must more often deduce receptor conformations from ligand efficacy than the other way around. We will pursue two goals: a. Leveraging its sole reliance on ligand structures, grouped by efficacy, we will adapt the Similarity Ensemble Approach (SEA) to differentiate groups of biased ligands. b. Further, we will predict and test new biased ligands at D2R for antipsychotic use. Milestones. The essential features of this system exist, and proof of principle has been demonstrated in predicting new targets for over 35 targets for 20 drugs using SEA. Here we refine SEA's notion of target to encode ligand bias. There are two pragmatic milestones. i. Evaluating five computational methods that leverage a ligand centric viewpoint to predict new biased ligands, from SEA alone to in house marker set receptor signatures. ii. Prospectively testing 20 arrestin biased D2R ligand predictions experimentally and advancing (a) one drug to antipsychotic animal models or (b) up to three preclinical compounds to PK profiling, as warranted. Aim 2: To relate GPCR pharmacology by biased ligands. We will extend our maps of all GPCRs to reflect biased ligand classes. This will allow us to interrogate the method's reach and impact: a. Determining whether arrestin biased ligands associate more closely in the map with canonical ligands of the same receptor-or instead with the arrestin biased ligands of an unrelated receptor. b. Adapting methods from Aim 1 to predict and test biased ligands for further therapeutic receptors. Milestones. We will relate functional selectivity to the wider context of known GPCR pharmacology by: i. Combining biased ligands from this study and from the literature into a global map of receptor relationships. ii. Demonstrating broad applicability by predicting and testing a further 20 biased ligands at each of two receptors (for 40 total) selected from the list f therapeutic receptors with ligands known to exhibit biased downstream signaling. Whereas these goals are admittedly ambitious, preliminary results suggest that they are feasible.